<rozzin>
Seems I'm not the only one--the one that's broken in that image, is the exact same one that I broke--in exactly the same way :)
<wolfspraul>
ah yes
<wolfspraul>
:-)
<wolfspraul>
they break easily, it's on the list of mechanical annoyances to do better in the next major hardware revision
<wolfspraul>
there is a total of 5 of them, they are not all really necessary because there are also 4 screws holding it down
<rozzin>
Most of them were just fine;
<wolfspraul>
but if the one next to the speaker breaks (like in that picture), you will probably see a bit wider opening between the case parts
<rozzin>
it's just the one in the corner that snapped,
<rozzin>
because the corner couldn't flex as well as the sides :)
<wolfspraul>
yeah. it will not impact functionality, only if you look there you will see the top and bottom case parts are not as tightly together as they should
<rozzin>
That's what I figured.
<rozzin>
There--all fixed.
<rozzin>
wolfspraul: On the whole, I think you guys did a pretty awesome job.
<wolfspraul>
thanks!
<wolfspraul>
we have ambitious plans, hopefully with all the support here we can make it...
<wolfspraul>
:-)
<wolfspraul>
thank you for buying a NanoNote!
<rozzin>
It's a *beautiful* device, both functionally and aesthetically,
<rozzin>
and I'm quite impressed by that kind of general `fit and finish' actually managing to coexist with the capacity of being easily disassembled *and* reassembled.
<rozzin>
That the only thing that broke in the process was that (mostly superfluous) corner latch.
<wolfspraul>
hmm
<wolfspraul>
we learnt a lot
<wolfspraul>
first there is a lot of glue still
<wolfspraul>
want to reduce that to zero next time
<wolfspraul>
top cover is glued on the piece beneath it
<wolfspraul>
"Qi Hardware" bezel is glued
<wolfspraul>
metal piece over speaker is glued
<wolfspraul>
small rubber feet are glued (and can fall off)
<wolfspraul>
those little latches around the edge are bad
<wolfspraul>
hinge cannot be tightened after use, so it will just wear out and that's it
<wolfspraul>
we'll get to all this :-)
<rozzin>
Fine by me.
<rozzin>
The most important parts are in good shape.
<wolfspraul>
the PC sheet over the LCD is also glued
<rozzin>
Not that I'm saying `here are some laurels, be complacent', of course :)
<wolfspraul>
but overall I also like the device
<wolfspraul>
I like to start with something real, start selling it
<wolfspraul>
that gives gravity to all other activities
<rozzin>
Yes indeed.
<wolfspraul>
it's easy to critizize things out of the air, when you do nothing
<wolfspraul>
but once you start, you realize everything has tradeoffs
<wolfspraul>
and improving the bottom line is actually hard :-)
<wolfspraul>
but that's the fun part
<wolfspraul>
and I do think the little Ben NanoNote is a worthwhile starting point...
<rozzin>
I actually like that it's as simplistic as it is--no radios or anything.
<rozzin>
I'm going to be sad if (when) you guys end up discontinuing Ben, down the road, in favour of more featureful models.
<rozzin>
Maybe I should stock up, now... :)
<wolfspraul>
rozzin: he. no worries. we are slow :-)
<rozzin>
I actually have some nice photographs to contribute to the disassembling-guide, now.
<rozzin>
That'll have to wait until tomorrow, though--it's far past bedtime, here.
<rozzin>
wolfspraul: Thanks for everthing :)
<wpwrak>
rozzin[zzz]: since you're already taking the ben apart, why not unsolder the microphone properly ? that way, you can attach it again in case you need it later.
<qwebirc15239>
Hi Wolfang, I'm Pietro and I'm ready to apply your instruction to 'resuscitate' my Nanonote
<wolfspraul>
hah
<wolfspraul>
qwebirc15239: cool
<wolfspraul>
what are your current symptoms?
<wolfspraul>
tried reflashing, now screen is always dark?
<wolfspraul>
I suggest we go straight to the usb-boot reflashing
<wolfspraul>
what do you think?
<qwebirc15239>
after the error I made (flashing it alway at the wrong offset  2048) it does not turn on, so I don't know how to flash it
<wolfspraul>
well there must have been some more
<wolfspraul>
but let's see
<wolfspraul>
are you ready?
<wolfspraul>
take out the battery
<qwebirc15239>
yes
<qwebirc15239>
done
<wolfspraul>
unplug the USB cable, leave it plugged into your computer
<qwebirc15239>
ok
<wolfspraul>
on your computer, run 'watch lsusb'
<wolfspraul>
close your nanonote, put it on the table upside down (empty battery compartment facing you)
<wolfspraul>
put the carbonized rubber button over the 'usb boot' pins, and press down firmly
<wolfspraul>
'firmly' means quite hard, no hesitation. just don't crush the device in hulk style...
<wolfspraul>
then, while you are holding the carbonized rubber button pressed down, plug the usb cable in from the side (you can leave the nanonote closed, no problem)
<wolfspraul>
keep holding the button down even 1-2 seconds after plugging the usb cable in
<wolfspraul>
then you can let the carbonized rubber button go (keep the usb cable plugged in), and look at your computer, the 'watch lsusb' output
<wolfspraul>
check whether you see a number 601a:4740 showing up
<qwebirc15239>
ok, you are a magician
<qwebirc15239>
now I can reflash it!!!
<wolfspraul>
well, let's see. first I want to see something on the lcm :-)
<wolfspraul>
we had a few cases with bad lcm recently
<wolfspraul>
so let's continue
<wolfspraul>
for now, only reflash u-boot
<wolfspraul>
let's do it manually
<wolfspraul>
on your computer, run 'usbboot'
<wolfspraul>
then you go into the usbboot console
<wolfspraul>
now type 'boot'
<wolfspraul>
then 'nerase 0 4096 0 0'
<wolfspraul>
do the 'boot' and 'nerase' commands run through?
<qwebirc15239>
run
<wolfspraul>
after the nerase, flash u-boot into page 0 "nprog 0 openwrt-xburst-qi_lb60-u-boot.bin 0 0 -n"
<wolfspraul>
if you are not in the directory where u-boot.bin is, exit usbboot, go to that directly, start usbboot again
<qwebirc15239>
yes
<wolfspraul>
no need to run the 'boot' and 'nerase' commands again
<wolfspraul>
btw, that nerase will erase the entire nand, so don't repeat it later when flashing the kernel or rootfs
<wolfspraul>
after you have flashed u-boot, exist usbboot and unplug the usb cable
<wolfspraul>
open your nanonote, plug in the usb cable
<wolfspraul>
you should see u-boot on the lcd, and it should complain about not finding a kernel
<qwebirc15239>
please wait, I'm in trouble exexcuting nprog
<wolfspraul>
sure
<qwebirc15239>
u-boot has been flashed, with " Finish! (len 454656 start_page 0 page_num 111)"; I made  unplug and plug again, but no reactions
<wolfspraul>
unplug the cable for 1 minute
<qwebirc15239>
ok, now I'm waiting
<wolfspraul>
:-)
<wolfspraul>
there is a hardware reset button too on the bottom of the device
<wolfspraul>
but let's just wait a bit
<wolfspraul>
after you plug in the usb cable, if it doesn't turn on you can also try to press the power-on button for several seconds, even 5
<wolfspraul>
if that still doesn't work, unplug the cable again, wait 10 seconds or so
<wolfspraul>
run 'watch lsusb' on your notebook
<wolfspraul>
hold down the 'u' button on the NanoNote, and keep it pressed, then plug in the usb cable
<wolfspraul>
see whether 0x601a:4740 shows up on the host
<qwebirc15239>
sorry, no reactions
<wpwrak>
wolfspraul: i wonder if you have a flowchart sitting next to your computer by now ;-)
<wolfspraul>
qwebirc15239: let's reflash the rest first
<wolfspraul>
unplug the usb cable
<wolfspraul>
run 'watch lsusb' on the host
<wolfspraul>
press the 'u' button on the nanonote, keep it pressed, plug in the usb cable
<wolfspraul>
do you see 601a:4740 on your host?
<qwebirc15239>
no
<wolfspraul>
alright
<wolfspraul>
then we flash with the usb-boot pins :-)
<wolfspraul>
unplug the usb cable
<wolfspraul>
close nanonote, turn upside down
<wolfspraul>
run 'watch lsusb' on host
<qwebirc15239>
done
<wolfspraul>
press carbonized rubber button over usb-boot pins, hold it there, plug usb cable in
<wolfspraul>
check whether 601a:4740 appears on host
<qwebirc15239>
appeared
<wolfspraul>
good
<wolfspraul>
by the way, why do you run the nprog commands manually?
<wolfspraul>
not that I like the script, but we have a reflash_ben.sh script that most people use
<wolfspraul>
in the email you sent me, you had the wrong offsets and I am wondering where they are coming from
<wolfspraul>
where did you download u-boot/kernel/rootfs from?
<wolfspraul>
maybe now that you have 601a:4740 on your host, you can try to just run reflash_ben.sh ?
<wolfspraul>
or do you want to run the nprog commands manually?
<qwebirc15239>
I tried it, but it stopped in error, informing me that the Nanonote was not replying something; rootfs have been dowloaded by reflash_ben.sh and I use those
<wolfspraul>
hmm
<wolfspraul>
ok then
<wolfspraul>
let's try manual commands
<wolfspraul>
go to the directory where you have u-boot, uImage and rootfs
<wolfspraul>
start 'usbboot'
<wolfspraul>
run 'boot'
<wolfspraul>
then 'nerase 0 4096 0 0'
<wpwrak>
wolfspraul: does reflash.sh download new images ? (idea: make it download a "known to be safe even if ancient" set of images by default. add some option or interactive query to download an update of itself, maybe with another name, that does download more current things. that way, you may have less trouble with the cases where nothing seems to work.)
<wolfspraul>
then 'nprog 0 u-boot.bin 0 0 -n'
<wolfspraul>
wpwrak: that script already has too much magic in it for me
<wpwrak>
wolfspraul: ;-))
<wolfspraul>
that's why I don't trust it
<wolfspraul>
qwebirc15239: then 'nprog 1024 uImage.bin 0 0 -n'
<wolfspraul>
finally 'nprog 2048 root.ubi 0 0 -n'
<wolfspraul>
(I shortened the filenames, I think it's clear what it should be)
<wolfspraul>
the nprog 2048 root.ubi might take 10-15 minutes
<qwebirc15239>
ok, I'll do all that and then I'll be back to the irc
<wpwrak>
wolfspraul: "son of the devirginator" then ;-) (not sure if you ever looked at that one. it was a bit before your time at OM, though they may have still used it)
<wolfspraul>
wait
<wolfspraul>
if you have another computer with Linux, try reflashing from there too
<wolfspraul>
first you say reflash_ben.sh failed with some error
<qwebirc15239>
ok
<wolfspraul>
then we just reflashed u-boot, but it seems didn't really arrive correctly in NAND
<wolfspraul>
otherwise at least the 'u' keypress should have worked
<wolfspraul>
to save time, you can focus on first flashing u-boot
<wolfspraul>
the screen must turn on, and u-boot will complain about not finding a kernel
<wolfspraul>
he, we can try my beloved xbboot as well
<wolfspraul>
are you up for a little experiment?
<qwebirc15239>
yes
<wolfspraul>
try running 'xbboot' on your computer
<qwebirc15239>
sudo xbboot openwrt-xburst-qi_lb60-zImage-initramfs.bin replies Error - unknown vendor request openwrt-xburst-qi_lb60-zImage-initramfs.bin - run with --help to see all requests
<wolfspraul>
:-)
<wolfspraul>
did you see the parameter I posted
<wolfspraul>
xbboot -u 0x80600000
<qwebirc15239>
ok I've seen now the instructions, sorry
<wolfspraul>
then the filename
<wolfspraul>
this will (should) load a little rescue system directly into the memory of your NanoNote, and execute it there
<wolfspraul>
well, if it works...
<wolfspraul>
what happens?
<qwebirc15239>
ok, bulk_write successfully wrote 3825664 bytes.
<wolfspraul>
do you see anything on the nanonote? (keep the usb cable plugged in)
<qwebirc15239>
no, dark; do I've to press power?
<wolfspraul>
no
<qwebirc15239>
now wathc lsusb shows
<wolfspraul>
what does it show?
<qwebirc15239>
Bus 001 Device 010: ID 0525:a4a2 Netchip Technology, Inc. Linux-USB Ethernet/RNDnIS Gadget
<wolfspraul>
that just appeared?
<qwebirc15239>
yes, after xbboot
<wolfspraul>
hmm. that's a strong indication that there is a hardware problem with your LCM.
<wolfspraul>
you need to send it in, we'll fix it. where did you buy the device?
<qwebirc15239>
but before my erros (wrong offsets) it was working; I started flashin to upload the graphic system and I made a disaster ...
<wolfspraul>
yes I understand
<qwebirc15239>
directly via web from you
<wolfspraul>
I am also surprised about how often a reflash results in (seemingly) LCM problems
<wolfspraul>
maybe it's not an LCM problem after all
<wolfspraul>
but in this xbboot case, it seems Linux really booted up
<qwebirc15239>
My idea: I'll redo in a calm way your instructions from the beginning and then I'll inform you
<wolfspraul>
but LCM stays dark? afaik that initramfs does enable the lcm
<wpwrak>
wolfspraul: could pressing hard on usb_boot trigger some latent mechanical problem ?
<wolfspraul>
you can try reflashing from another computer
<qwebirc15239>
yes
<wolfspraul>
I am wondering why your original reflash failed.
<wolfspraul>
qwebirc15239: where did you buy the device?
<qwebirc15239>
I don't press too hard
<wolfspraul>
wpwrak: don't know
<wpwrak>
qwebirc15239: maybe try the whole flashing process, with u-boot, kernel, and rootfs, and see how this goes.
<qwebirc15239>
yes, in the next few hours; then I'll send you an email; many many thanks for your help
<wolfspraul>
no problem thanks for your patience
<wolfspraul>
we'll get it fixed again, no worries
<qwebirc15239>
:-)
<wolfspraul>
and you get more confident reflashing stuff, also nice
<wolfspraul>
:-)
<wolfspraul>
qwebirc15239: where did you buy the device?
<wolfspraul>
qwebirc15239: btw, you don't need to hurry. there is no warranty tricks or anything. you have time. if you cannot get your nanonote back alive, we will fix or replace it.
<qwebirc15239>
I bought it directly from qi-hadrware, via web; I'll try to solve the problem and thean I'll be back to you
<wolfspraul>
ah OK, which country are you in?
<qwebirc15239>
Italy
<wolfspraul>
and you bought from our Hong Kong online shop?
<wolfspraul>
maybe from Tuxbrain
<wolfspraul>
we'll figure it out...
<qwebirc15239>
my confrimation emails come from hackable-devices.org
<qwebirc15239>
see you later
<qwebirc13861>
how can I access serial port of Nanonote to connect with my PC
<wpwrak>
wolfspraul: btw, the webirc instructions should probably include a suggestion to use /nick. it does get a bit confusing ...
<wpwrak>
wolfspraul: btw2, have you actually posted to the list about how the 2nd 1k run went ?
<wolfspraul>
wpwrak: not yet, have to write the mail...
<wolfspraul>
actually if we change the url to not use 'randomnick' then people have to choose a nick first
<wolfspraul>
it's a tradeoff between '1-click' and something that is better in the long run
<wolfspraul>
I figure the barrier of entry should be as low as possible, hence the url with the randomnick
<wpwrak>
(randomnick) ah, that's why. well, as i said, it does get confusing with several qwebircs showing up regularly
<wolfspraul>
sure
<wolfspraul>
but like I said, you add that one additional form and many people will not make it here at all :-)
<wolfspraul>
you can ignore that or explain it with 'their' lack of whatever, or you lower the barrier so that they do show up here
<wpwrak>
also, in a way, it dehumanizes them, making others more likely to ignore them
<wolfspraul>
raise the barrier of entry and they will not show up at all
<wolfspraul>
of course they still exist, just you don't see them anymore
<wpwrak>
well, one extra question shouldn't be *that* evil :)
<wolfspraul>
we should do a test with 100 randomly chosen people from the streets of Buenos Aires
<wpwrak>
or is the nick choice particularly awkward ?
<wolfspraul>
you would be surprised about the loss
<wolfspraul>
no just another form
<wolfspraul>
of course you better not choose a nick that is already in use or registered at freenode, then you probably get more forms
<wolfspraul>
we could display a welcome message, I think that would be helpful
<wolfspraul>
some people click on that link, but then they don't know what to do
<wpwrak>
hmm, considering that having to sign up for any kind of forum is already the standard experience, i'm not so sure it should be overly confusing
<wpwrak>
also, you can add a troubleshooting mail address next to the IRC instructions. see how much you get :)
<wolfspraul>
wpwrak: the run went well
<wolfspraul>
the only outstanding problem is a weird USB bug, maybe on the electrical side, maybe in the 4720 silicon
<wolfspraul>
long story
<wpwrak>
oh, a new bug ?
<wolfspraul>
like last time, we chose to just ignore this bug and discard the board
<wolfspraul>
it's a long story
<wolfspraul>
production bugs first of all typically only show on some devices
<wolfspraul>
that makes it a production bug
<wolfspraul>
last time we had a device where the rightmost column of keys wouldn't work
<wpwrak>
"rinse and repeat" goes a long way. particularly at such small numbers ;)
<wolfspraul>
but they did work with the proprietary software
<wpwrak>
heh ;-)
<wolfspraul>
1 out of 1000, and it did not work with our software
<wolfspraul>
that's normal
<wolfspraul>
this time I have a device that issues a 'interrupt 80' flood
<wolfspraul>
took it out, send to xiangfu
<wolfspraul>
but no time to study right now
<wolfspraul>
who knows could be anything
<wolfspraul>
hardware is not binary
<wolfspraul>
as you know
<wolfspraul>
everything has tolerances, nothing is 100% precise
<wolfspraul>
the bottom line is that I'm happy :-)
<wolfspraul>
it was obvious that our software got a lot better
<wolfspraul>
much less strange NAND behavior
<wolfspraul>
maybe I had 5 devices or so where flashing caused an error, and just redoing the whole flashing from scratch would fix it
<wolfspraul>
on the first run that was more like 50 :-)
<wolfspraul>
the flashing throughput was also increased, everything got a little better
<wolfspraul>
last time 3 days, 280+320+360
<wolfspraul>
this time on the first day we already flashed 640, and started only at about 3 PM
<wolfspraul>
I have many ideas now how to improve things more...
<wolfspraul>
many little bugs here and there
<wolfspraul>
1000 fully tested and flashed nanos are being shipped to HK this week, any day now
<wpwrak>
with 1k/day you're good to handle a lot, particularly if you can switch from wolfgang-days to chinese-factory-worker-days :)
<wolfspraul>
what is a wolfgang-day and how does it differ from a chinese-factory-worker-day?
<wolfspraul>
I take the liberty to typically show up a bit late, they cannot :-)
<wpwrak>
the latter are available in larger quantity and less valuable :)
<wolfspraul>
but on the other end they easily beat me, crazy as it is. stay until 2-3 AM, one time even 6.30 AM
<wpwrak>
not bad :)
<wolfspraul>
ah yes, it's already at the point where they could do it
<wolfspraul>
but I like to see reality once in a while
<wolfspraul>
nothing beats hands-on experience
<wolfspraul>
also on the testing side, I have many ideas now how to automate this whole thing
<wolfspraul>
actually Linux is quite suitable for it
<wolfspraul>
good scripting
<wolfspraul>
can output audio over the speaker, record it over the microphone, compare, etc.
<wolfspraul>
that's for later...
<wpwrak>
yup. also, doing it yourself helps to identify problems the factory may never report
<wpwrak>
and automation is lovely for sure :)
<wolfspraul>
wpwrak: ah, for the ben-wpan RF lab. I'm up for the challenge. If you think it's ready, I suggest you email what you want to have tested to the list.
<wolfspraul>
then you need to ship one board to Adam, or actually we can even try for him to make some already, if you think the chances of this being already somewhat usable by others are high enough.
<wolfspraul>
of course if he needs to make them, it will slow things down
<wolfspraul>
then we prepare a little testing setup, some scripts, and Adam goes to an RF lab and we see what we can get out of it
<wpwrak>
kewl. i'm currently investigating what rf lab options i have around here. i'd like to at least tune the antenna first. that affects overall physical size and all that.
<wolfspraul>
ok
<wolfspraul>
most important for me (or Adam) is to know what you actually want to have tested
<wolfspraul>
but writing that up should be a good thing anyhow
<wpwrak>
i'm less worried about spectrum byproducts. picking the right components should largely avoid them. and i'm now pretty conservative about them. i want to switch baluns, though. the one i'm currently using is quite expensive. (a custom design for this transceiver, made in germany)
<wolfspraul>
right now all I know is "Werner wants to have the antenna tested", but that's a little weak on the details :-)
<wpwrak>
there's one that's about half the price, still specifically for this chip, from a us company
<wolfspraul>
we can also look for some Chinese or Asian baluns?
<wolfspraul>
if Chinese is scary (it is to me), then maybe Korean or Japanese could be a good compromise
<wpwrak>
the tricky bit about the antenna is that it's test, adjust (new pcb), test, etc.
<wpwrak>
those baluns are optimized for this chip. don't now if there are any from other sources. probably not.
<wolfspraul>
hmm
<wpwrak>
there are "generaL" baluns but then need more rf tuning. also, with the extra components, they're probably not even much cheaper.
<wolfspraul>
what is a 'balun' after all?
<wolfspraul>
what are the biggest drivers of cost on the board right now?
<wpwrak>
it adapts the differential output of the chip to the antenna
<wolfspraul>
how much are all components roughly, right now (without pcb)?
<wolfspraul>
the rf ic is 2.20 USD, the balun 67 cents
<wpwrak>
i may have to tweak some caps. the ones on the list are okay according to what atmel specify, but then the spec may not tell the whole truth. so maybe ten cents per unit more.
<wpwrak>
i like the price :)
<wpwrak>
there's one more item i may add. that would be a demultiplexer for a control signal that i'm currently ignoring
<wolfspraul>
well to me the atmel is still an 'expensive' chip, but definitely this is a great starting point
<wolfspraul>
and unlike Ron thinks, we will not start our own home-grown RF ICs now...
<wolfspraul>
that 67 cents balun is already the cheaper one you are looking at, or the more expensive one?
<wpwrak>
in theory, the signal could be important. in practice, i'm less sure. first of all, it may be possible to do completely without it and just tweak via some other signals. second, we may not be able to use the functionality it controls (very precise timing) anyway.
<wpwrak>
(that is, not without a dedicated mcu with very tight response time and such)
<wolfspraul>
ah you wrote it already, sorry
<wolfspraul>
so the other one is 1.30 USD
<wpwrak>
(home-grown) naw, ron got that right
<wolfspraul>
sure if it works switching to that cheaper balun makes sense
<wpwrak>
it's actually more. just checked. USD 1.82750@100
<wpwrak>
the circuit between chip and antenna is the single-chip balun in one case (plus two capacitors and a just-in-case resistor)
<wpwrak>
in the other, it's six components for balun plus filter
<wpwrak>
(the dc blocking cap at the end in atusb, C3, probably isn't necessary)
<wolfspraul>
hmm, I see
<wpwrak>
so the integrated balun removes quite a bit of stuff. stuff that needs tuning and such. tuning with very high-end equipment.
<wolfspraul>
wpwrak: so you are saying maybe you will stay with the 1.87 USD balun in the end?
<wolfspraul>
also OK, don't kill yourself
<wpwrak>
pov-ray can sometimes be funny. the aspect ratio changes when you change the size of the projection area. clever. so just cropping is harder than one would expect. and, of course, it explains why i was getting strange distortions all the time :)
<wpwrak>
no, i want the cheap balun
<wpwrak>
but i don't want to go back to a set of discrete components, like in the first version of atusb
<wpwrak>
the cheap balun is on my shopping list :) wouldn't want to expect it to work without actually testing it. though it should be fine "out of the box".
<wolfspraul>
oh I see. those 2 schematics are not expensive balun vs. cheap balun, but expensive balun vs. self-made balun?///
<wpwrak>
yup. the cheap one only became available at digi-key a few days ago, so i first tried to avoid the expensive one. but then i realized that this made the design just too messy.
<wolfspraul>
got it
<wpwrak>
also considering the limitations of my pcb process. i.e., i can't make particularly good vias.
<wpwrak>
industrial pcb makers will have kittens with the shape of the board ;-)
<wpwrak>
so .. with wpan, my schedule is to a) order the cheap balun, and b) try to get some antenna testing and adjustment done around here. once all this is done, then adam can have fun with it
<wpwrak>
i probably won't order the cheap balun this week but wait until i have another item to add. but that one needs a test circuit, which in turn needs the mill. so there's a bit of a dependency chain.
<wpwrak>
but i can move ahead with the antenna testing without waiting for the balun
<wpwrak>
now it's better. the lighting still suck, though.
<xiangfu_away>
qwebirc13861: I don't know much about Max232. there should be some output when NanoNote boot.
<qwebirc13861>
there is a ttl output on Tx and Rx points while using minicom but during conversion from TTL to 232 no output
<qwebirc13861>
xiangfu_away> : who can guide me for this
<xiangfu_away>
qwebirc13861: what you mean 'during conversion' .
<qwebirc13861>
at TTL to 232 converter level
<wpwrak>
qwebirc13861: is the "ttl" (3.3 V) signal still there when you connect the converter ? does the converter actually work (without the ben) ?
<xiangfu_away>
need offline for dinner. see you later
<wolfspraul>
wpwrak: [ben-bottom] is there a shadow on some parts?
<wolfspraul>
looks good though
<wpwrak>
yes, lots of shadows. and the ambient light (to make the shadows go away) doesn't work too well. if i make it stronger, i don't get much shadow at all.
<wpwrak>
there's proabably a way to make light just penetrate objects. so each would look as if rendered separately, sort of an engineering view.
<wpwrak>
or just make then semi-transparent, the usual ray tracing show-off kind of thing :)
<wpwrak>
but one thing after the other. first, i want to beat solidify into a shape that makes it usable for others
<wpwrak>
wolfspraul: btw, the plan for the case scans is to do the pcbs now, and then battery. not quite sure what to do about the lcd. i don't have an lcd without pcb. one option is to just make a rectangular block. another option would be to scan the pcb+lcd combo (i already have half of it), subtract the pcb, and see if this comes out right.
<wpwrak>
wolfspraul: same thing with the sponge. i'm not sure if i can do much with the smaller components, cover lock, led rubber, and such. they're difficult to mount.
<wpwrak>
wolfspraul: the black and clear plastic that goes in front of the lcd will have to be constructed. first, i don't have it as a scannable part (all the glue gets in the way), and second, its main characteristic - being partially clear - would be lost.
<wpwrak>
wolfspraul: also the metal bits (top of the shell, speaker) may need the same process as lcd and such. only that they're more difficult. so i'm not sure if this will work. worst case, they would have to be replaced with constructed parts, too.
<wpwrak>
wolfspraul: once i have the whole collection of parts in a displayable way, i'll post thickness measurements. with them and the scans, anyone can try their luck at aligning faces and maybe compensating for scan problems or distortions of the part.
<wpwrak>
wolfspraul: finally, we'll have to find a way to get solids a cad system can use. right now, all i have is an output pov-ray likes. someone who understands something about how 3d solids are implemented in cad systems would be useful for taking care of this. i.e., the kind of "brute force" approach i used with pov-ray may not work so well with real cad systems.
<wpwrak>
wolfspraul: then, these solids with ragged surfaces would have to be translated into properly engineered parts ...
<wpwrak>
of course, one could also use pov-ray as a poor man's cad for such things. show one of the parts like i have them now, then make the constructed one (in pov-ray), and let pov-ray display the difference between the two. crude, but it may work.
<wpwrak>
eventually, we'll run into prototyping. right now, there are two active players: kristianpaul who has a 3d printer and me with my mill. the 3d printer is amazingly fast and it can do full 3d shapes relatively easily. not sure about overhangs. one its problem is print quality, both in terms of overall accuracy and in terms of surface finish.
<wpwrak>
my mill can make very precise and smooth surfaces, but it's not as fast (even extremely slow for some things) and it's not so good for parts that have two faces, because i need to flip and part, which introduces significant variations along the z axis.
<wpwrak>
one way around this problem is to make molds instead of directly milling parts. in a mold, only one face matters. the problem then becomes what material to cast. i now can do, but lead isn't the best choice for everything ;-) i've started to investigate plastics and wax+plastic combinations, but it's still uncertain if anything useful will come out of this.
<kristianpaul>
wpwrak: i can get .3 mm per layer right now is higher that that i think .375
<kristianpaul>
wpwrak: i can try add support material by software wich means more plastic in the printin (easy to remove), right now overhangs that are more than 45 degrees are a bit challenging ;)
<qwebirc13861>
anybody worked upon serial port of nanonote ?
<wpwrak>
kristianpaul: doens't help without usb host on the ben
<kristianpaul>
qwebirc13861: no, just the always question "why you need a MCU there? "
<kristianpaul>
wpwrak: ahh i tought was just with usb client
<wpwrak>
kristianpaul: i think we wants the serial-usb only for development. but i could be wrong :)
<kristianpaul>
i mean ben may behave as an usb cable using already ethernet client capabillties
<kristianpaul>
hmm i was thinking other thing
<kristianpaul>
but it seems :)
<qwebirc13861>
ok
<qwebirc13861>
now tell me about TTL level serial port
<qwebirc13861>
how can I go for it
<kristianpaul>
ok
<wpwrak>
calling it "ethernet" is a bit misleading in this context
<kristianpaul>
jej
<wpwrak>
ttl to the external mcu ? first, make sure that one does have the same levels. btw, "ttl" is normally considered to mean 5 V. the ben uses 3.3 V.
<wpwrak>
almost all modern electronics of this kind use 3.3 V, of course. so you're likely to be compatible.
<kristianpaul>
some tll support 3.3 byt joinin gor opening a jumper
<wolfspraul>
we should embed them in a nice wiki page with instructions
<wolfspraul>
I must have missed them, looking now...
<wpwrak>
so, good enough to get a feeling for the part, but not really prototype quality. not sure if this can be improved, e.g., by working the surface with a light solvent
<wpwrak>
that's the kind of accuracy issue this sort of printer has
<wolfspraul>
yeah but it's nice
<wolfspraul>
I don't think we have to copy whatever is currently the 'standard'
<wpwrak>
oh yes, it's pretty darn cool what they can do
<wpwrak>
the problem is that you need ~0.5 mm structures and ~0.1 mm accuracy fairly quickly
<wolfspraul>
we market it as organic/recycled computers :-)
<wpwrak>
even for simple parts, engineered for easy making
<wpwrak>
heh ;)
<wpwrak>
well, you could use wood ... :)
<wolfspraul>
oh I'll do that
<wolfspraul>
picture with 2 apples, iPhone, our device
<wolfspraul>
iPhone together with waxed & shiny cheap apple
<wolfspraul>
our device with natural, worm & bird bitten organic Apple
<wpwrak>
maybe put a few dead worms around the Apple apple :)
<wolfspraul>
yeah, exactly
<wpwrak>
wood could be fun. you can treat it with PUR and then it resists pretty much anything. not quite sure about swelling, though.
<wpwrak>
of course, one problem with wood is that it would need machining from at least two sides.
<wolfspraul>
my main concern is cost / unit
<wolfspraul>
I don't want to make expensive toys
<wolfspraul>
unfortunately we are up against a big challenge on the cost side
<wpwrak>
i don't think case making has to be expensive
<wpwrak>
injection molding of small quantities (a few kunits) should be relatively inexpensive if you do everything up to making the mold
<wpwrak>
what's harder are the prototypes. they have to be reasonably accurate.
<wpwrak>
ah, regarding kristianpaul's 3d prints: they have a lot of right angles. you can get away with this with a 3d printer, but a part made for molding would have to have round edges. so if you 3d-print such a design, the result may be better, too.
<wpwrak>
(with a mill, outer corners can be as pointy as you want but inner corners are round, too)
<wpwrak>
aah, finally it rains ! hay fever, be gone !
<rafa>
wpwrak: ;-))
<rafa>
larsc: you there? did you modify that driver for the SD wifi in qi?
<rafa>
larsc: I am modifying it for 2.6.34.. a lot of changes to do ! :(
<larsc>
rafa: i think it worked out of the box on 2.6.32
<rafa>
larsc: well.. that tar.gz has a patch for the kernel and the sources to build the module
<rafa>
the patch is okey, but the sources of the driver to build the module needs a lot of changes for 2.6.34 ..
<rafa>
larsc: but well, if you did not do many changes then I will continue with the work. I was thinking that if you needed to do a lot of changes then maybe you alredy did the ones I need to do :)
<rafa>
larsc: no, I do not know how to find stuff, yet, inside of that whole xburst tree :(
<rafa>
larsc: let me check, thanks to point
<rafa>
larsc: I am checking.. that driver has the same modifications I did :P
<rafa>
larsc: and it does not work "out of the box".. or maybe for you, to modify around 300 lines means "works out of the box" ;-) for me no, I need to check every api which changed and to modify
<kristianpaul>
wolfspraul: about printings, yes is not a wiki page yet, i think i can create one to night
<kristianpaul>
is just a lazy first print i i think i can get better resolution in Z axis just need do more tests
<kristianpaul>
wpwrak: had you tried make round angles in small parts in Heekscad?
<kristianpaul>
wpwrak: my printer is a makerbot cupcake cnc if i dint tell you before btw
<kristianpaul>
wpwrak: so talking about the case how you said it still fat as in the last git commint, but why?
<kristianpaul>
what i need i prove? :)
<wpwrak>
(coveyor) wow, cool :)
<tuxbrain>
coveyor
<tuxbrain>
?
<kristianpaul>
lol
<wpwrak>
kristianpaul: (rounding) that would be fillet and chamfer. the rightmost items in heekscad's top menu bar
<wpwrak>
kristianpaul: and no, i haven't used them yet
<kristianpaul>
wpwrak: i'll try
<kristianpaul>
wpwrak: what about the fatness?
<wpwrak>
kristianpaul: rounding the edges will probably make them print in a nicer way
<kristianpaul>
also help in overhang
<wpwrak>
(fat) thin is sexy ;-)
<wpwrak>
(overhang) yup, that too
<kristianpaul>
wpwrak: how thin is sexy?
<wpwrak>
and such a design is compatible with a lot more production technologies
<wpwrak>
0.5 mm of air below the board. 1 mm above.
<kristianpaul>
show me more
<kristianpaul>
show me a pic or something similar you have in mind
<kristianpaul>
ah 0.5 thats challeing
<kristianpaul>
challenging*
<wpwrak>
i don't have a picture :) just thinner
<wpwrak>
why is 0.5 mm a problem ?
<kristianpaul>
ok
<kristianpaul>
is not
<wpwrak>
ok :)
<wpwrak>
(acid) you mean for smoothing ? naw, the abs may resist that. but some pastic solvents (paint thinner or such) may help to soften the abs a little. needs experimenting, though.
<kristianpaul>
yeah thinner will
<kristianpaul>
let me find a cheap way of send stuff to argetnina too
<wpwrak>
"ABS polymers are resistant to aqueous acids, alkalis, concentrated hydrochloric and phosphoric acids, alcohols and animal, vegetable and mineral oils, "
<wpwrak>
[...] "They are soluble in esters, ketones and ethylene dichloride."
<wpwrak>
"It is degraded (dissolved) when exposed to acetone."Â Â that one's easy to get :)
<kristianpaul>
i think i have some acetone
<kristianpaul>
it had to used before to clean the machine nozzle before
<kristianpaul>
after i discover firing it is more fun and efective :D
<kristianpaul>
okay i-ll remove other 3mm to the design
<kristianpaul>
is that okay for you?
<wpwrak>
oh. not problems with carbonized residues ?
<wpwrak>
s/not/no/
<kristianpaul>
well not that i noticed
<kristianpaul>
was a small cilinder
<wpwrak>
3 mm less sounds great :) how tall is it then in total ? (top and bottom part)
<kristianpaul>
right now is 6mm
<kristianpaul>
but i dont have heeks here to cofirm
<kristianpaul>
you said the wpan is 3mm in height?
<wpwrak>
hmm, 6mm - 3 mm would be 3 mm. that sounds too thin. youi have bottom thickness, top thichness, 0.5 mm air at bottom, ~1 mm air above pcb at top, 0.8 mm of pcb.
<wpwrak>
a bit less than 3 mm, yes. around 2.3-2.5 mm.
<kristianpaul>
see !
<kristianpaul>
or
<kristianpaul>
i can make tall the part it is and ther other stuff thin as much i can
<kristianpaul>
but i will not look well i think
<wpwrak>
i have one that measures 2.15 mm. so around 2.3-2.5 mm would provide for some tolerances
<kristianpaul>
ok
<kristianpaul>
so 4.5mm (top-botton) is thin and sexy?
<kristianpaul>
plus i-ll try round edges
<kristianpaul>
hope that dont increase the fatness
<wpwrak>
4.5 mm total sounds good
<wpwrak>
that way, it stays a little below the edge of the ben
<kristianpaul>
hehe we can swap the board and solve that problem :p
<wpwrak>
hmm, that would actually be possible :) more vias, though
<wpwrak>
of cours, then the ben would probably stand on the box :)
<kristianpaul>
hmm
<rozzin>
Hm. Is there a serial number somewhere on Ben NanoNote?
<larsc>
nope
<rozzin>
Interesting.
<larsc>
why?
<rozzin>
Well, it causes difficulties in situations where organisations want to maintain a registry of devices.
<rozzin>
Unless there's some other unique ID that can be substituted.
<rozzin>
Like, I don't think that I could buy one for business use at any company where I've worked.
<rozzin>
If it's an untrackable asset.
<larsc>
put a sticker on it
<rozzin>
That's not how the corporate policies have worked.
<rozzin>
As far as I know, anyway
<mth>
depends on the company: where I used to work, they put their own stickers with serial number on everything, regardless of the manufacturer serial numbers
<mth>
probably because their own stickers could be scanned by barcode
<mth>
(on everything of value: a monitor would have a sticker but a mouse would not)
<rozzin>
Actually, I may be wrong:
<rozzin>
I think the mice that I had my current employer get me lacked serial numbers.
<rozzin>
The electronic computer-accessory kind of mice, not the biological kind, of course....
<rozzin>
Though, I presume not-having-serial-numbers isn't an issue in places that use the animals.
<mth>
well, if you're testing medicine on mice then I think it would be a very good idea to give them serial numbers, or you might draw the wrong conclusions
<rozzin>
Well, yeah--I presume they'd get asset-numbers or something.
<rozzin>
But they wouldn't come with manufacturer serial numbers :)
<wpwrak>
rozzin: no DNA ? :)
<rafa>
larsc: I managed to build the wifi driver for 2.6.34 kernel.. but no luck yet. when I insmod it It complains that "ks79xx_sdio: Unknown symbol wireless_send_event"
<rafa>
larsc: I traced a bit.. wireless_send_event is defined in net/wireless/wext-core.c
<rafa>
and it EXPORT_SYMBOL that
<rafa>
the 2.6.34 kernel is set to build the kernel wireless code and I also
<rafa>
read wireless_send_event in System.map after kernel built..
<rafa>
but no idea yet how to continue.. thinking and checking google :)
<wpwrak>
rafa: is it in /proc/kallsyms ?
<rafa>
wpwrak: let me check..
<wpwrak>
rafa: you can just  grep wireless /proc/kallsyms
<rafa>
wpwrak: ahh. cool.. nop :(
<rafa>
it is not there
<wpwrak>
rafa: there usually aren't a lot of things (~20)
<rafa>
wpwrak: no idea how to add that now :)
<rafa>
I would guess some option in .config.. I have checked with menuconfig and I have all it would need
<wpwrak>
rafa: is anything else wireless there ?
<rafa>
nothing
<wpwrak>
rafa: erm, basics first: are you sure you're running the kernel you built (the one that has wireless in System.map) ?
<wpwrak>
rafa: wouldn't be the first time this sort of mishap happens ;-) (you can check .version)
<rafa>
well, yes.. if bootloader works how it says that it works :)
<rafa>
I can rm all the uImage under /boot/ and just put my last kernel built.. which is now as /boot/uImage
<mth>
comparing .version to the output of uname -v might be easier
<wpwrak>
i was more thinking of copying the wrong kernel or not copying it (due to a typo or such), not necessarily a boot loader bug
<rafa>
wpwrak: ah.. no.. wait.. funny.. I have the kernel in sd and driver.ko in nand.. soooooooo
<rafa>
I am running the nand kernel. HA!
<wpwrak>
*grin*
<rafa>
:D
<rafa>
thanks.. let me reboot this thing
<rozzin>
wolfspraul: I do actually have one complaint about Ben NanoNote.
<wolfspraul>
shoot
<wolfspraul>
(btw, for the record, that Italian guy who had a dark LCM yesterday fully unbricked his NanoNote now, he sent me a mail...)
<wolfspraul>
I'm wondering what happened to tugasoft
<wpwrak>
wolfspraul: what was the problem with the LCM ?
<rozzin>
The accessibility of the colour-coding on the keys is poor.